Abstract
The use in analysis of variance of six different ways of calculating the Land Equivalent Ratio (LER) was assessed. Criteria used are (1) normality of the residuals of LER values after fitting block and treatment effects, (2) precision of comparisons arising from the analysis of variance, and (3) possible bias in the means. These were examined in detail for data from one experiment and consistency of the results was investigated for seven other sets of data. Separate standardization in each block had no advantages over using the same standardization in all blocks. The use of many different divisors can lead to problems in the statistical analysis of LERs.